Sign in to follow this  
mikemariano

Dreaming The Simulation Dream

Recommended Posts

Rock, Paper, Shotgun's most recent Sunday Papers linked to a Gamasutra article called The Simulation Dream.

 

Tynan Sylvester's article argues that the results of a simulation need to be exposed and meaningful to the player.  Even further, the impression that a player can make something happen is more important than actually modelling that specific system! (As a player, he had a jealous Sim hatch a murder plot, even though The Sims doesn't specifically model jealousy or revenge.)

 

What really blew my mind were the blog entries designer Raph Koster linked to in the comments: Areas in which NPCs can be improved and three separate articles about how resources were originally meant to be handled in Ultima Online.

 

These are articles from 2006 about design documents from 1995!  I want to believe that we can create better simulations now.  I want to believe they can make better games.

 

Have any of you played games that you felt could use more background simulation that would have added to your experience as a player?  Or are there any games you can think of where the simulation bits got in the way?

Share this post


Link to post
Share on other sites

What really blew my mind were the blog entries designer Raph Koster linked to in the comments: Areas in which NPCs can be improved and three separate articles about how resources were originally meant to be handled in Ultima Online.

 

Someone should link to those articles on the Project Eternity and Wasteland kickstarters if they haven't already. I'd hate to think they'll still be relevant.

Share this post


Link to post
Share on other sites

I enjoy the idea of tuning games to evoke more apophenia. Tarot cards are one of the best examples of this in my opinion. I think that tarot is especially effective for a few reasons. First of all, its got a great amount of hype that makes the player very keen on interpretation. The "hair-complexity" that the author recommends is the details, desires, and concerns of the player's own life. Many forms of divination do this, especially the ones that still have a mystique and respect surrounding them. Another thing tarot does well is that it starts with "story-rich" archtypes that can be infered with very little knowledge. This is more true for the major arcana, but the story-richness of that subset of the deck is so concentrated with concepts like "secrets" and "lovers", that the player ends up being invested enough by seeing some of those cards that they are willing to look in a book to find out what the 3 of swords reversed means.

The tarot is pretty much the epitome of what the author is talking about IMO.

One option that i don't see the author considering is just giving the player a path to see the complexity in a simulation heavy game. I have certainly played games where it seems like a lot is going on, but it all just looks kinda dull and repeatitive. But I've also found ways to appreciate the complex systems.

I've mentioned Artificial Life for ios in another thread, but it's such a good example. The "game" is hundreds of two-dimensional NPC micro-organisms sliding around in a petri dish with their own hierachy of behaviors. The super cool part is that when they mate, both parents contribute 6(?) behaviors into their offspring which has a total of 12(?) behaviors.

When I play the game, it ends up looking like a bunch of creatures running around meaninglessly, but there are waysto appreciate it. One way is to follow an individual NPC for a while. At that scope, the "hair-compexity" would be somewhat valuable, but ultimately novel. Following around individual protozoa gets boring.

The Artificial Life "game" suggests that programming an individual NPC with your own instructions and watching to see how long it survives is a good strategy for appreciating the complexity, but it doesn't help the player appreciate the complexity as much as it helps them appreciate the potential.

The way I've come to enjoy the "game" is to observed various areas of the petri dish over time. Once the simulation has been running for a while, evolutionary strategies emmerge and are incredibly interesting. I've seen colonies who do things like hibernation until they become weak, and only then di they feed or mate. Other colonies will swarm around the petri dish just ravaging anything that they are not related to. Some colonies will basically become herbivores, depending on high fertility and maintaining a proximity to where food spawns. The challenge for the designer is to communicate objecties and provide interactive capabilities that will quickly lead the player to discover and invest in the complexity.

I think that in the case of Artificia Life, a good way of doing this would be to give the playerthe ability cordone off areas and move individual protozoa into protected areas. This would allow players to breed the protozoa.

Another way to encourage the player to discover and appreciate the complexity of a simulation is to allow he player to create "fitness functions" for the genetic algorithm. I hope that behaviors will be exchanged between prisoners in Prison Architect (similar to Artificial Life's mating mechanic) because I expect that the player will be able to create fitness functions for prisoners by conditioning with things like regular food, comfort, and socialization.

The main point of the gamasutra article is a good one. Designers can't just make complex simulations and expect that players will be able to appreciate them. Designers also have to consider the "Player model" and how they will expose the interesting simulation aspects to the player. I understand the idea that apophenia should be prioritized over unseen complex systems, but I think that finding ways to expose the player to thecomplex systems in a meaningful way is also a good option.

Share this post


Link to post
Share on other sites

i like all kinds of simulations in games and the more the merrier in my opinion, however i don't think pure simulation always works well, i think there should always be a safety net in the simulation that doesn't allow the game to become broken, like a game where you could hunt an animal to extinction would be bad, there should always be a hidden magic couple that breed and don't allow extinction because what if just by starting the simulation you end up with a non sustainable ecosystem, in real life evolution and adaptation happens making a generally balanced self correcting ecosystem, but in a game if you programme the antelope AI and the lion AI for specific behaviours the antelope won't adapt or change their behavior because the lions started the simulation in an advantageous position and the lion AI won't change accordingly the lions will just kill all the antelope and then they will all die of starvation or just move on to the next prey, you could fix that with even more complex simulation, but until we get to the holodeck level of simulation there will always need to be some kind of safety net.

 

generally i want to see the results of the simulation else there isn't much point in doing it, and also sometimes complex simulation ends up just making something simple, like in prison architect the developers are talking about the prisoners as individuals with their own personal (hidden) traits but i am playing a game with like 40+ "individuals" in it but they just become a crowd so their individuality is gone, it's the same with dwarf fortress, each dwarf has an insane level of traits and individual preferences, but because i play the game from a big picture perspective their individual needs and preferences are meaningless they are just part of a big crowd that has crowd needs and preferences, i guess that is more a psychology thing than simulation thing but I'm a man not a machine so the psychology of the player should be something to concentrate on along side the simulation 

Share this post


Link to post
Share on other sites

Imagine if in Prison Architect you could hire some sort of Internal Affairs agent to investigate events such as prisoner murders, guard abuse, or riots. If that agent could then use a game-log to create a report that consists of a chain of causality, then the individual traits and preferences of the prisoners could be shown to have purpose. If i got a report back that a riot started because two prisoners got in a fight over a smuggled cigarette and they had some sort of gang alliances, that would imply meaning to the simulation.

I think this is the type of thing these games need. Spacebase DF-9 was inspired with their social app attempt, but ultimately it didn't seem to actually reflect much from the simulations details. I suspect that the information may have to be shown in a causality chain in order to have meaning to the player (unless you want to depend on apophenia and the player's imagination).

Share this post


Link to post
Share on other sites

maybe that would help with prison architect but unless you can individually deal with prisoners they would still just be part of the crowd, and the way i see it is if i have to imagine it i may aswell not play it and just entirely imagine it because my imagination will always beat the game, i think this whole idea of story creation in a simulation is just a way for journalist to fill pages not an actual feature in a game, i like to experiment and mess around with mechanics and maybe set my own goals which some may describe as a story but i wouldn't 

Share this post


Link to post
Share on other sites

Well, to make it apply to the crowd, you could get a weekly incident report. Everytime someone gets shived in the yard, it makes a note. After a while you start to see a pattern and fundamental attribution error falls away. You can then hire the internal investigator or interview a variety of individual prisoners who were involved in the incidents that seem to occur in the same place. You notice that most of the causality chains involve not being able to get through a door or a toilet or something. Then you as a player can change the architecture to influence behavior of the crowd. And it's because you had feedback from various scopes.

Share this post


Link to post
Share on other sites

Well, to make it apply to the crowd, you could get a weekly incident report. Everytime someone gets shived in the yard, it makes a note. After a while you start to see a pattern and fundamental attribution error falls away. You can then hire the internal investigator or interview a variety of individual prisoners who were involved in the incidents that seem to occur in the same place. You notice that most of the causality chains involve not being able to get through a door or a toilet or something. Then you as a player can change the architecture to influence behavior of the crowd. And it's because you had feedback from various scopes.

 

well i don't know if changing the architecture affects their behaviour (except if a room is so far away they can't get to it in time with their schedule) but if it did that would be useful feedback, i think useful feedback would tell you that generally they crowd needs more entertainment or free time etc. would be good

Share this post


Link to post
Share on other sites

What really blew my mind were the blog entries designer Raph Koster linked to in the comments: Areas in which NPCs can be improved and three separate articles about how resources were originally meant to be handled in Ultima Online.

 

These are articles from 2006 about design documents from 1995!  I want to believe that we can create better simulations now.  I want to believe they can make better games.

 

Thanks for linking that, those were a good read! Interesting what a simple generic resource system could achieve. Of course, seems they never proved that it would work well, and admitted the problem of unseen causality.

 

Has anyone who worked on the Stalker ALife systems written about them?

Share this post


Link to post
Share on other sites

in real life evolution and adaptation happens making a generally balanced self correcting ecosystem, but in a game if you programme the antelope AI and the lion AI for specific behaviours the antelope won't adapt or change their behavior because the lions started the simulation in an advantageous position and the lion AI won't change accordingly the lions will just kill all the antelope and then they will all die of starvation or just move on to the next prey, you could fix that with even more complex simulation, but until we get to the holodeck level of simulation there will always need to be some kind of safety net.

 

Kinda off-topic but, I wonder, has there ever been a game where the environment evolves to 'counter' the player's actions? Not specifically an actual nature/animal simulation, but any game where the AI - or even the levels themselves - change dynamically depending on how you play. Valve's AI Director is similar to what I mean, but unless I'm mistaken that doesn't actually modify level design or enemy AI, only spawning and whatnot (I'm too cool to google these things).

If something like that does exist, and isn't terrible, I'm just gonna start throwing money at the screen.

Share this post


Link to post
Share on other sites

well i don't know if changing the architecture affects their behaviour (except if a room is so far away they can't get to it in time with their schedule) but if it did that would be useful feedback, i think useful feedback would tell you that generally they crowd needs more entertainment or free time etc. would be good

I suppose that one of the things that having a micro and macro feedback layer could do is imply that the circumstance of the individual and the circumstance of the group, are both valuable. Maybe the circumstances of one NPC within the group is always a symptom of a larger circumstance. So if little Jonnie isn't getting her medication on time, then it's a sign that the pharmacy is under-supplied.

Share this post


Link to post
Share on other sites

Kinda off-topic but, I wonder, has there ever been a game where the environment evolves to 'counter' the player's actions? Not specifically an actual nature/animal simulation, but any game where the AI - or even the levels themselves - change dynamically depending on how you play. Valve's AI Director is similar to what I mean, but unless I'm mistaken that doesn't actually modify level design or enemy AI, only spawning and whatnot (I'm too cool to google these things).

If something like that does exist, and isn't terrible, I'm just gonna start throwing money at the screen.

yeah that does sound like a good idea, but from my experience AI is one of those things that hasn't changed in the last ten years,

but yeah i can think of a few types of games where an evolving level would be very cool, like a platformer type game where there are multiple types of ways to progress forwards that use different techniques (maybe multiple tools like grappling hooks or portal guns and just plain old jumping) but the more you use one type of way to progress forwards the level evolves to make those techniques harder so you would have to keep doing things differently,

 

or an RTS/tower defence game where each wave tries to counter the base you have built or different enemies spawn on the map depending on how strong your fortress is, or a shooter where there are procedurally generated AI creatures that evolve depending on how you kill them and how long it takes for you to kill them eg. one creature could evolve armour on its head because you keep headshotting it or another creature could develop faster more erratic behaviour to dodge more bullets etc.

Share this post


Link to post
Share on other sites

Oh dude. It would be so cool if the enemies you have a harder time killing have a chance to reproduce and bequeath their armored heads or tendency to run away from plasma weapons.

Share this post


Link to post
Share on other sites

Oh dude. It would be so cool if the enemies you have a harder time killing have a chance to reproduce and bequeath their armored heads or tendency to run away from plasma weapons.

it could be a soldier testing chamber run by GLaDOS's army AI brother or some kind of hell dimension with a similar antagonist to GLaDOS where they keep building/spawning new evolving enemies at you depending on how you kill them. man "procedurally generated evolving AI creatures" is something someone has to do, imagine an online multiplayer hoard mode like that where it would be different each time you played, developers you can have that idea for free, infact i will pay you if you make that game

Share this post


Link to post
Share on other sites

Okay, I'm getting a shitload of ideas now, I'm going to go code something before my head explodes (or this is announced as Half-Life: Episode 3)

Share this post


Link to post
Share on other sites

Open world games suffer especially from lack of simulation or lack of story to make simulation meaningful. The two polar examples are: Mount and Blade relies on simulation too much, making it feel more samey the longer you play, while Skyrim is story-driven, and its simulation is very thin. Once its thin simulation layer is pierced, Skyrim feels like a very hollow experience to me. 

Share this post


Link to post
Share on other sites

Aren't there some Skyrim quests that put you into a different alignment for other quests? I did a quest in Markath that got me in trouble with the law and I ran away thinking "Well I can't ever come back here." Then the main quest line was like "Meet you in Markath." I figured that a statute of limitations or main quest prioritization might have cleared my name. Nope.

I would certainly prefer more simulation effects like that to happen in Skyrim. I like how the dragons seem to be willing to attack you in any scenario.

Going back to the original linked article, Skyrim has a lot of great "hair-complexity" with the race relations, establishment of symbolic relationships between NPC's and their ownership of homes and such. But I agree with you, I would love the status quo in Skyrim to break more frequently just by the method I use to complete a quest rather than having to go on a killing spree to really see some changes.

I think you are right about Mount&Blade also. I wonder if the game could add enough story just by increasing the quantity of Mayor missions and things that generals and villagers ask you to do. I haven't played much Fire&Sword because I HATE the guns. You can't just ride up into a crowd of enemies and start hacking like you could in the old days. Times have changed and the battlefield with it.

Share this post


Link to post
Share on other sites

Going back to the original linked article, Skyrim has a lot of great "hair-complexity" with the race relations, establishment of symbolic relationships between NPC's and their ownership of homes and such. But I agree with you, I would love the status quo in Skyrim to break more frequently just by the method I use to complete a quest rather than having to go on a killing spree to really see some changes.

 

 

In Fallout: New Vegas, you start to get attacked by random groups of NCR Rangers or Legionnaires, depending on how you've aligned your character.  They seem to spawn at hardcoded spots on the map, deep enough into each other's territory to make it absurd from a story perspective, but also far enough away from opposing outposts, animal enemies, etc. to make it uninteresting from a systems perspective.  It's always disappointing when a game has the opportunity to let its systems clash (lure the Legion marauders to an NCR outpost!) but intentionally shy away from it.

 

Similarly, in Deus Ex: Human Revolution, you can fight police and civilians in the hub areas, but every mission area is strictly Jensen vs. commandos.  There is a mission specifically about civilians getting killed, but you only gain control of Jensen after they are all dead.  Perhaps their AI is only reliable when they are hunting you?

 

Games are afraid of their own systems.

Share this post


Link to post
Share on other sites

Well, to make it apply to the crowd, you could get a weekly incident report. Everytime someone gets shived in the yard, it makes a note. After a while you start to see a pattern and fundamental attribution error falls away. You can then hire the internal investigator or interview a variety of individual prisoners who were involved in the incidents that seem to occur in the same place. You notice that most of the causality chains involve not being able to get through a door or a toilet or something. Then you as a player can change the architecture to influence behavior of the crowd. And it's because you had feedback from various scopes.

 

I can think of several games off the top of my head (Empire: Total War, Miasmata, Sim City) where the clouding of the game mechanics was hiding an underlying lack of depth in the game. In all three of those cases, I was enjoying the game greatly when I was assuming that there were tons of comlicated gears turning behind the scenes but, when I found out that it was infact one old man on a bicycle, I enjoyed the game a lot less. For Total War and Sim City, I stopped playing the game entirely. For Miasmata, I had already finished the game (and had fun) but was a bit dissapointed to read FAQs after. I get the feeling from what I've seen of Prison Architect, it might not be quite as deep as it's trying to seem.

 

I guess what I'm trying to say is that hiding how some of those mechanics work, though a bit dishonest, can make a game seem more interesting. Obviously, I'd like to just see more machines with complicated systems (like X3) but, at times where that's not possible, I'm not adverse to a little misdirection.

Share this post


Link to post
Share on other sites

So I'm reading Pride And Prejudice and I have found that I really enjoy Austen's style. Today, I was appreciating something about her writing that reminded me of the problems posed in this thread, not the solutions. I'm going to tell you about what I read, and then explain how it demonstrates a solution for appreciation of complex systems in games.

 

Here is an excerpt from Chapter 1 [spoilers I suppose]

Wickham told Elizabeth that Mr. Darcy is a dick and providing an account of Mr.
Darcy being dickish. Elizabeth is now telling her sister Jane, who is courting
Mr. Darcy's best-friend, so Jane is somewhat invested in Mr. Darcy not being
perceived as a dick.

 

Elizabeth related to Jane the next day, what had passed between Mr. Wickham and herself. Jane listened with astonishment and concern; -- she knew not how to believe that Mr. Darcy could be so unworthy of Mr. Bingley's regard; and yet, it was not in her nature to question the veracity of a young man of such amiable appearance as Wickham. -- The possibility of his having really endured such unkindness, was enough to interest all her tender feelings; and nothing therefore remained to be done, but to think well of them both, to defend the conduct of each, and throw into the account of accident or mistake, whatever could not be otherwise explained.
``They have both,'' said she, ``been deceived, I dare say, in some way or other, of which we can form no idea. Interested people have perhaps misrepresented each to the other. It is, in short, impossible for us to conjecture the causes or circumstances which may have alienated them, without actual blame on either side.''

 

``Very true, indeed; -- and now, my dear Jane, what have you got to say in behalf of the interested people who have probably been concerned in the business? -- Do clear them too, or we shall be obliged to think ill of somebody.''

 

Alright, here is what I like about Austen's style here. She explains the complex reasoning for Jane's opinion before using dialogue to show Jane's opinion. If Austen had skipped the reasoning, and just written the dialogue, then it wouldn't have engaged me. Wickham has an amiable appearance, so Jane defends him when attacked; Mr. Darcy is a friend of Bingley, so she defends him when he is attacked. Both of them are being attacked, both must be defended; so Jane suggests that there must be a misunderstanding. This conditional behavior is established, and then through the dialogue, the reader sees it played out. I think this style of mapping out the why and then watching the instance of how is psychologically satisfying. The satisfying process is to understand a system, predict behavior with the system, and witness a demonstration.
 

Ok, here is where I think this relevant to this thread and the article in the OP that I just re-read. The article does a great job of showing what is valuable about prioritizing apophenia in a game, but I'm skeptical that the original problem it poses is caused by having a complex system. I think the problem is that the player doesn't notice, or even have access to the rules of the complex system. If the systems within the game are more apparent, then we can really appreciate them when they play out. NPC's in Ultima could have reported low populations of certain species to players and detailed three direct chains of causation for why the animal was in low population and what would happen if it stays low or gets higher. Of course I would prefer this to be somewhat elegant, but you get the idea, tell the player how the system interacts within the game world. This information will give the player a way to play with the system rather than just affect it.  

 



 


 

Share this post


Link to post
Share on other sites

NPC's in Ultima could have reported low populations of certain species to players and detailed three direct chains of causation for why the animal was in low population and what would happen if it stays low or gets higher.

 

"I didn't see a mudcrab the other day.  Someone must have killed them all."

 

That's a great example you got from Pride and Prejudice.  Austen may have been a little blunt in exposing her character motivations in that first paragraph, but she makes it pay off with that joke.  I do wonder, though, if exposing parts of the Ultima Online systems would reward the players trying to roleplay more than it would get exploited by min/maxers and gold farmers.

Share this post


Link to post
Share on other sites

Have you ever heard of the "cobra effect"

http://en.m.wikipedia.org/wiki/Cobra_effect

An npc faction could offer bounties that would make erradication unprofitable. Then to make sure that they never go completely extinct, they could introduce migrations from the borders after everyone has learned their lesson. I just see so much interesting potential in all this. It would be so cool if there was a moon-cycle and certain species would migrate into the world on during certain moon-phases, like sea-turtles.

I think that if the simulation was designed in a way that expects min-maxers, then they can use their greed and efficiency in favor of the simulation.

The article in the OP seems to Intentionally give examples of failed ecosystems. One is an early MMO where the designers had no concept of how crowds act, and the other is a game that never actually implimented the simulated ecosystem. I understand that it is easier to prioritize apophenia and not create an interactive simulation, but I don't think that such a focus has the potential to maintain interest while the details of how the "simulation" works are revealed through play. It's like Dewar said in eesaldrom's post, it's enjoyable when you think there is some complexity to the gears, but when you find out it's a simple trick, the player loses interest.

I think that there is a lot of potential in games where simulated ecosystems interact and players get a stream of info or experiences that shows them what the rules of those interactions are.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this